TickerTain — Make Analysis Fun Again

▶ 15:52 · $HARVEY · Harvey builds custom legal AI harness; LangChain ran SLM judge experiment with them

now playing · $HARVEY

LangChain CEO presents agent harness architecture and automated eval platform

Harrison Chase argues that owning the agent harness — the orchestration layer connecting models, context, and tools — is essential for compounding AI intelligence, and demonstrate…

$LANGCHAINneutral· highharrison chase

LangChain CEO unveils Deep Agents and LangSmith Engine to automate agent improvement flywheel

LangChain provides a minimal base harness (LangChain) and customizable Deep Agents with model-aware middleware, while LangSmith Engine automates the trace-to-fix loop to compound…

$HARVEYbullish· mediumharrison chase

Harvey builds custom legal AI harness; LangChain ran SLM judge experiment with them

Harrison cites Harvey as a prime example of a domain-specific company building its own custom harness for legal AI, and reveals LangChain collaborated with Harvey to fine-tune sma…

$ANTHROPICneutral· mediumharrison chase

Anthropic's Claude Code harness uses distinct file-editing approach vs OpenAI

Harrison notes that Anthropic and OpenAI have converged on strong coding capabilities but landed on different file-editing implementations in their harnesses (Claude Code vs Codex…

$OPENAIneutral· mediumharrison chase

OpenAI's Codex harness uses distinct file-editing approach vs Anthropic

Harrison notes that OpenAI and Anthropic have converged on strong coding capabilities but landed on different file-editing implementations in their harnesses (Codex vs Claude Code…

←